tensor: Add FP16 parameter and expose to VX_core

hw/rtl/VX_core_wrapper.sv

@@ -4,6 +4,7 @@
 
 module Vortex import VX_gpu_pkg::*; #(
     parameter CORE_ID = 0,
+    parameter TENSOR_FP16 = 0,
     parameter BOOTROM_HANG100 = 32'h10100,
     parameter NUM_THREADS = 0
 ) (
@@ -394,7 +395,8 @@ module Vortex import VX_gpu_pkg::*; #(
 
     // TODO: SCOPE_IO_BIND should be socket id
     VX_core #(
-        .CORE_ID (CORE_ID)
+        .CORE_ID (CORE_ID),
+        .TENSOR_FP16 (TENSOR_FP16)
     ) core (
         `SCOPE_IO_BIND  (0)
 

hw/rtl/core/VX_core.sv

@@ -18,7 +18,8 @@
 `endif
 
 module VX_core import VX_gpu_pkg::*; #( 
-    parameter CORE_ID = 0
+    parameter CORE_ID = 0,
+    parameter TENSOR_FP16 = 0
 ) (        
     `SCOPE_IO_DECL
     
@@ -191,7 +192,8 @@ module VX_core import VX_gpu_pkg::*; #(
     );
 
     VX_execute #(
-        .CORE_ID (CORE_ID)
+        .CORE_ID (CORE_ID),
+        .TENSOR_FP16 (TENSOR_FP16)
     ) execute (
         `SCOPE_IO_BIND  (2)
         

hw/rtl/core/VX_execute.sv

@@ -14,7 +14,8 @@
 `include "VX_define.vh"
 
 module VX_execute import VX_gpu_pkg::*; #(
-    parameter CORE_ID = 0
+    parameter CORE_ID = 0,
+    parameter TENSOR_FP16 = 0
 ) (
     `SCOPE_IO_DECL
 
@@ -144,7 +145,7 @@ module VX_execute import VX_gpu_pkg::*; #(
 
 `ifdef EXT_T_ENABLE
     VX_tensor_core #(
-        
+        .FP16 (TENSOR_FP16)
     ) tensor_core (
         .clk(clk),
         .reset(reset),

hw/rtl/core/VX_tensor_core.sv

@@ -2,7 +2,7 @@
 `include "VX_fpu_define.vh"
 
 module VX_tensor_core import VX_gpu_pkg::*; #(
-
+    parameter FP16
 ) (
     input clk,
     input reset,
@@ -52,7 +52,8 @@ module VX_tensor_core import VX_gpu_pkg::*; #(
 
     for (genvar block_idx = 0; block_idx < BLOCK_SIZE; ++block_idx) begin
         VX_tensor_core_block #(
-            .ISW(1) // FIXME: not block_idx
+            .ISW(1), // FIXME: not block_idx
+            .FP16(FP16)
         ) tensor_core (
             .clk(clk),
             .reset(reset),
@@ -65,7 +66,8 @@ module VX_tensor_core import VX_gpu_pkg::*; #(
 endmodule
 
 module VX_tensor_core_block import VX_gpu_pkg::*; #(
-    parameter ISW
+    parameter ISW,
+    parameter FP16
 ) (
     input clk,
     input reset,
@@ -121,7 +123,8 @@ module VX_tensor_core_block import VX_gpu_pkg::*; #(
 
         VX_tensor_octet #(
             .ISW(ISW),
-            .OCTET(i)
+            .OCTET(i),
+            .FP16(FP16)
         ) octet (
             .clk(clk),
             .reset(reset),
@@ -329,6 +332,7 @@ endmodule
 module VX_tensor_octet #(
     parameter ISW,
     parameter OCTET,
+    parameter FP16,
     parameter RESULT_BUFFER_DEPTH = 2
 ) (
     input clk,
@@ -519,6 +523,7 @@ module VX_tensor_octet #(
     VX_tensor_threadgroups #(
         .ISW(ISW),
         .OCTET(OCTET),
+        .FP16(FP16),
         .OPERAND_BUFFER_DEPTH(4 /*@perf: arbtirary*/)
     ) dpu (
         .clk(clk),

hw/rtl/fpu/VX_tensor_dpu.sv

@@ -5,6 +5,7 @@
 module VX_tensor_threadgroups #(
     parameter ISW,
     parameter OCTET,
+    parameter FP16,
     // @perf: has big impact on throughput.  A rule of thumb is to set it to
     // the pipeline length of FEDPs in order to make sure there are enough
     // entries to fully saturate the pipeline, but this is still rough
@@ -102,6 +103,7 @@ module VX_tensor_threadgroups #(
     // threadgroup DPUs; B_tile is shared across the two threadgroups. See
     // Figure 13 in paper
     VX_tensor_threadgroup #(
+        .FP16(FP16)
     ) threadgroup_0 (
         .clk   (clk),
         .reset (reset),
@@ -115,6 +117,7 @@ module VX_tensor_threadgroups #(
         .D_frag    (D_tile[1:0])
     );
     VX_tensor_threadgroup #(
+        .FP16(FP16)
     ) threadgroup_1 (
         .clk   (clk),
         .reset (reset),
@@ -165,7 +168,7 @@ endmodule
 // does (m,n,k) = (2,4,2) matmul compute over 2 cycles.
 // see Figure 10(b) of the paper.
 module VX_tensor_threadgroup #(
-    parameter HALF_PRECISION = 1
+    parameter FP16
 ) (
     input clk,
     input reset,
@@ -297,7 +300,7 @@ module VX_tensor_threadgroup #(
         wire [31:0] d_col_sel = (substep_in == 1'b0) ? d_col : (d_col + 1);
 
         // Dot product (FEDP) unit generated from Chisel
-        if (HALF_PRECISION != 0) begin
+        if (FP16 != 0) begin
             TensorDotProductUnit fedp (
               .clock (clk),
               .reset (reset),